Analytification and Tropicalization Over Non-Archimedean Fields

ANALYTIFICATION AND TROPICALIZATION OVER NON-ARCHIMEDEAN FIELDS

ANNETTE WERNER

Abstract:

In this paper, we provide an overview of recent progress on the interplay between tropical geometry and non-archimedean analytic geometry in the sense of Berkovich. After brieﬂy discussing results by Baker, Payne and Rabinoﬀ [BPR11] [BPR13] in the case of curves, we explain a result from [CHW14] comparing the tropical Grassmannian of planes to the analytic Grassmannian. We also give an overview of most of the results in [GRW14], where a general higher-dimensional theory is developed. In particular, we explain the construction of generalized skeleta in [GRW14] which are polyhedral substructures of Berkovich spaces lending themselves to comparison with tropicalizations. We discuss the slope formula for the valuation of rational functions and explain two results on the comparison between polyhedral substructures of Berkovich spaces and tropicalizations.

2010 MSC: 14G22, 14T05

1. Introduction

Tropical varieties are polyhedral images of varieties over non-archimedean ﬁelds. They are obtained by applying the valuation map to a set of toric coordinates. From the very beginning, analytic geometry was present in the systematic study of tropical varieties, e.g. in [EKL06] where rigid analytic varieties are used. The theory of Berkovich spaces which leads to spaces with nicer topological properties than rigid varieties is even better suited to study tropicalizations.

A result of Payne [Pay09] states that the Berkovich space associated to an algebraic variety is homeomorphic to the inverse limit of all tropicalizations in toric varieties. However, individual tropicalizations may fail to capture topological features of an analytic space.

arXiv:1506.04846v1 [math.AG] 16 Jun 2015 The present paper is an overview of recent results on the relationship between analytic spaces and tropicalizations. In particular, we address the question of whether a given tropicalization is contained in a Berkovich space as a combinatorial substructure.

A novel feature of Berkovich spaces compared to rigid analytic varieties is that they contain interesting piecewise linear combinatorial structures. In fact, Berkovich curves are, very roughly speaking, generalized graphs, where inﬁnite ramiﬁcations along a dense set of points is allowed, see [Ber90], chapter 4, [BaRu10] and [BPR13]. 1 2 ANNETTE WERNER

In higher dimensions, the structure of Berkovich analytic spaces is more involved, but still they often contain piecewise linear substructures as deformation retracts. This was a crucial tool in Berkovich’s proof of local contractibility for smooth analytic spaces, see [Ber99] and [Ber04]. Berkovich constructs these piecewise linear substructures as so-called skeleta of suitable models (or ﬁbration of models). These skeleta basically capture the incidence structure of the irreducible components in the special ﬁber.

In dimension one, Baker, Payne and Rabinoﬀ studied the relationship between tropicalizations and subgraphs of Berkovich curves in [BPR11] and [BPR13]. Their results show that every ﬁnite subgraph of a Berkovich curve admits a faithful, i.e. homeomorphic and isometric tropicalization. They also prove that every tropicalization with tropical multiplicity one everywhere is isometric to a subgraph of the Berkovich curve.

As a ﬁrst higher dimensional example, the Grassmannian of planes was studied in [CHW14]. The tropical Grassmannian of planes has an interesting combinatorial structure and is a moduli space for phylogenetic trees. It is shown in [CHW14] that it is homeomorphic to a closed subset of the Berkovich analytic Grassmannian.

In [GRW14], the higher dimensional situation is analyzed from a general point of view. This approach is based on a generalized notion of a Berkovich skeleton which is associated to the datum of a semistable model plus a horizontal divisor. This naturally leads to unbounded skeleta and generalizes a well-known construction on curves, see [Tyo12] and [BPR13].

For every rational function f with support in the fixed horizontal divisor it is shown in [GRW14] that the “tropicalization” log |f| factors through a piecewise linear function on the generalized skeleton. This function satisfies a slope formula which is a kind of balancing condition around any 1-codimensional polyhedral face. Moreover, for every generalized skeleton there exists a faithful tropicalization – where faithful in higher dimension refers to the preservation of the integral affine structures. In dimension one, this can be expressed via metrics. It is also proven that tropicalizations with tropical multiplicity one everywhere admit sections of the tropicalization map, which generalizes the above-mentioned result in [BPR11] on curves.

The paper is organized as follows. In section 2 we collect basic facts on Berkovich spaces and tropicalizations, giving references to the literature for proofs and more details. In section 3 we briefly recall some of the results by Baker, Payne and Rabinoff [BPR11] on curves. Section 4 starts with the definition and basic properties of the tropical Grass- mannian. Theorem 4.1 claims the existence of a continuous section to the tropicalization map on the projective tropical Grassmannian. We give a sketch of the proof for the dense torus orbit, where some constructions are easier to explain than in the general case. In section 5 we explain the construction of generalized skeleta from [GRW14], and in section 6 we investigate log |f| for rational functions f. In particular, Theorem 6.5 states the slope formula. Section 7 explains the faithful tropicalization results in higher dimension. ANALYTIFICATION AND TROPICALIZATION OVER NON-ARCHIMEDEAN FIELDS 3

Acknowledgements: The author is very grateful to Walter Gubler for his helpful com- ments.

2. Berkovich spaces and tropicalizations

2.1. Notation and conventions. A non-archimedean field is a field with a non-archimedean absolute value. Our ground field is a non-archimedean field which is complete with respect to its absolute value. Examples are the field Qp, which is the completion of Q after the p-adic absolute value, finite extensions of Qp and also the p-adic cousin Cp of the complex numbers which is defined as the completion of the algebraic closure of Qp. The field of formal Laurent series k((t)) over an arbitrary base field k is another example. Besides, we can endow any field k with the trivial absolute value (which is one on all non-zero elements). A nice feature of Berkovich’s general approach is that it also gives an interesting theory in the case of a trivially valued field. The reason behind this is that Berkovich geometry encompasses also points with values in transcendental field extensions – and those may well carry interesting non-trivial valuations.

Let K be a complete non-archimedean field. We write K◦ = {x ∈ K : |x| ≤ 1} for the ring of integers in K, and K◦◦ = {x ∈ K : |x| < 1} for the valuation ideal. The quotient K˜ = K◦/K◦◦ is the residue field of K. The valuation on K× associated to the absolute × value is given by v(x) = − log |x|. By Γ = v(K ) ⊂ R we denote the value group. A variety over K is an irreducible, reduced and separated scheme of finite type over K.

2.2. Berkovich spaces. Let us brieﬂy recall some basic results about Berkovich spaces. This theory was developed in the ground-breaking treatise [Ber90]. The survey papers [Co08], [Du07] and [Tem] provide additional information. For background information on non-archimedean ﬁelds and Banach algebras and for an account of rigid analytic geometry see [BGR84] and [Bo14].

For n ∈ N und every n−tuple r = (r1, . . . , rn) of positive real numbers we deﬁne the associated Tate algebra as −1 X I I K{r x} = { aI x : |aI |r → 0 as |I| → ∞}. n I=(i1,...,in)∈N0

I i in Here we put x = (x1, . . . , xn), and for I = (i1, . . . , in) we write x = x 1 . . . x . Moreover, we deﬁne |I| = i1 + ... + in.

A Banach algebra A is called K-affinoid, if there exists a surjective K-algebra homomorphism α : K{r−1x} → A for some n and r such that the Banach norm on A is equivalent the quotient seminorm induced by α. If such an epimorphism can be found with r = (1,..., 1), then A is called strictly K-affinoid. In rigid analytic geometry only strictly K-affinoid algebras are considered (and called affinoid algebras). 4 ANNETTE WERNER

The Berkovich spectrum M(A) of an aﬃnoid algebras is deﬁned as the set of all bounded (by the Banach norm), multiplicative seminorms on A. It is endowed with the coarsest topology such that all evaluation maps on functions in A are continuous.

If x is a seminorm on an algebra A, and f ∈ A, we follow the usual notational convention and write |f(x)| for x(f), i.e. for the real number which we get by evaluating the seminorm x on f.

The Shilov boundary of a Berkovich spectrum M(A) of a K-aﬃnoid algebra A is the unique minimal subset Γ (with respect to inclusion) such that for every f ∈ A the evaluation map M(A) → R≥0 given by x 7→ |f(x)| attains its maximum on Γ. Hence for every f ∈ A there exists a point z in the Shilov boundary such that |f(z)| ≥ |f(x)| for all x ∈ M(A). The Shilov boundary of M(A) exists and is a ﬁnite set by [Ber90], Corollary 2.4.5.

The Berkovich spectrum of a K-affinoid algebra carries a sheaf of analytic functions which we will not define here. Some care is needed since, very roughly speaking, not all coverings are suitable for glueing analytic functions. Analytic spaces over K are ringed spaces which are locally modelled on Berkovich spectra of affinoid algebras.

There is a GAGA-functor, associating to every K-scheme X locally of finite type an analytic space Xan over K. It has the property that X is connected, separated over K or proper over K, repectively, if and only if the topological space Xan is arcwise connected, Hausdorff, or compact, respectively. If X = SpecR is affine, the topological space Xan can be identifed with the set of all multiplicative seminorms on the coordinate ring R extending the absolute value on K. This space is endowed with the coarsest topology such that for all f ∈ R the evaluation map on f is continuous.

With the help of the GAGA-functor we associate analytic spaces to algebraic varieties. Another way of obtaining analytic spaces is via admissible formal schemes. An admissible formal scheme over K◦ is, roughly speaking, a formal scheme over K◦ such that the formal ◦ ◦ affine building blocks are given by K -flat algebras of the form K {x1, . . . , xn}/a for a ◦ P I ◦ finitely generated ideal a in the ring K {x1, . . . , xn} = { n aI x : aI ∈ K and |aI | → I∈N0 0 as |I| → ∞}. For a precise definition see [Bo14] or [Co08].

An admissible formal scheme X has a natural analytic generic fiber Xη. On the formal ◦ affine building block given by K {x1, . . . , xn}/a the analytic generic fiber is the Berkovich spectrum of K{x1, . . . , xn}/aK{x1, . . . , xn}.

If we start with a K◦- scheme X , which is separated, flat and of finite presentation, its completion with respect to any element π ∈ K◦◦\{0} is an admissible formal scheme X . This has an analytic generic fiber Xη. On the other hand, we can apply the GAGA-functor an to the algebraic generic fiber X = X ⊗K◦ K and get another analytic space X . They are an ◦ connected by a morphism Xη ,→ X , which is an isomorphism if X is proper over K . ◦ 1 an In the basic example X = SpecK [x] we get the natural inclusion M(K{x}) ,→ (AK ) , ANALYTIFICATION AND TROPICALIZATION OVER NON-ARCHIMEDEAN FIELDS 5 whose image is the set of all multiplicative seminorms on the polynomial ring K[x] which extend the absolute value on K and whose value on x is bounded by 1.

±1 ±1 2.3. Tropicalization. We start by considering a split torus T = SpecK[x1 , . . . , xn ]. an n Evaluating seminorms on characters we get a map trop : T → R on the associated Berkovich space, which is given by

trop(p) = (− log |x1(p)|,..., − log |xn(p)|). Here we are using the valuation map to deﬁne tropicalizations. In some papers, tropicalizations are deﬁned via the negative valuation map, i.e. the logarithmic absolute value of the coordinates.

If X is a variety over K together with a closed embedding ϕ : X,→ T , we consider the composition an an ϕ an trop n (2.1) tropϕ : X −→ T −→ R of the tropicalization map with the embedding ϕ. Note that the map tropϕ is continuous. an n Its image Tropϕ(X) = tropϕ(X ) is the support of a polyhedral complex Σ in R . This complex is integral Γ-aﬃne in the sense of section 5.1. It has pure dimension d = dim(X). For a nice introductory text on tropical geometry see the textbook [Ma-St]. The survey paper [Gu13] is also very useful.

−1 Note that for all ω ∈ Tropϕ(X) the preimage tropϕ ({ω}) can be identiﬁed with the Berkovich spectrum of an aﬃnoid algebra, see [Gu07], Proposition 4.1.

For every point ω ∈ Tropϕ(X) there is an associated initial degeneration, which is the special fiber of a model of X over the valuation ring in a suitable non-archimedean extension field of K. The geometric number of irreducible components of this special fiber is the tropical multiplicity of ω. There is a balancing formula for tropicalizations involv- ing tropical multiplicities on all maximal-dimensional faces of Tropϕ(X) around a fixed codimension one face, see [Ma-St], chapter 3.4.

Let K be the algebraic closure of K. It can be endowed with an absolute value extending n the one on K. By means of the coordinates x1, . . . , xn we can identify T (K) = K . n Then we can deﬁne a natural tropicalization map trop : T (K) → R by trop(t1, . . . , tn) = (− log |t1|,..., − log |tn|). Note that every point t = (t1, . . . , tn) in T (K) gives rise to the ±1 ±1 multiplicative seminorm f 7→ |f(t)| on the coordinate ring K[x1 , . . . , xn ], which is an element in T an. Hence the tropicalization map on T (K) is induced by the map trop on T an.

We could deﬁne the tropicalization Tropϕ(X) without recourse to Berkovich spaces. If the absolute value on K is non-trivial, the tropicalization Tropϕ(X) is equal to the closure of the image of the map ϕ trop n X(K) −→ T (K) −→ R . 6 ANNETTE WERNER

Considering a tropical variety as the image of an analytic space makes some topological considerations easier. For example, in this way it is evident that the tropicalization of a connected variety is connected since it is a continuous image of a connected space.

Let Y be a toric variety associated to the fan ∆ in NR, where N is the cocharacter group of the dense torus T . Then ∆ defines a natural partial compactification N ∆ (due to R Kashiwara and Payne) of the space NR. Roughly speaking, we compactify cones by dual spaces. For details see [Pay09], section 3. There is also a natural tropicalization map an ∆ trop : Y → NR , which extends the tropicalization map on the dense torus. As an example, we consider n n n Y = PK with its dense torus Gm,K and N = Z . Put R = R ∪ {∞} and endow it with the natural topology such that half-open intervalls ]a, ∞] form a basis of open neighbourhoods n of ∞. Then the associated compactification of NR = R is the tropical projective space n n+1 TP = R r {(∞,..., ∞)} /R(1,..., 1) which is endowed with the product-quotient topology. The Berkovich projective space n an n+1 an (PK ) can be described as the set of equivalence classes of elements in (AK ) \{0} n+1 an with respect to the following equivalence relation: Let x and y be points in (AK ) \{0}, i.e. multiplicative seminorms on K[x0, . . . , xn] extending the absolute value on K. They are equivalent if and only if there exists a constant c > 0 such that for every homogeneous polynomial f of degree d we have |f(y)| = cd|f(x)|.

n We can describe the tropicalization map on the toric variety PK as follows: n an n (2.2) trop : (PK ) → TP maps the class of the seminorm p on K[x0, . . . , xn] to the class (− log |x0(p)|,..., − log |xn(p)|)+ R(1,..., 1) in tropical projective space. An important result about the relation between tropical and analytic geometry due to Payne says that for every quasi-projective variety X over K the associated analytic space an X is homeomorphic to the inverse limit over all Tropϕ(X), where the limit runs over all all closed embeddings ϕ : X,→ Y in a quasi-projective toric variety Y (see [Pay09], Theorem 4.2). For a generalization which omits the quasi-projectivity hypothesis see [FGP14]. Hence tropicalizations are combinatorial images of analytic spaces, which recover the full analytic space in the projective limit. Individual tropicalizations however may not faithfully depict all topological features of the analytic space. We will discuss a basic example at the beginning of the next section.

3. The case of curves

Let K be a ﬁeld which is algebraically closed and complete with respect to a non- archimedean, non-trivial absolute value, and let X be a smooth curve over K. ANALYTIFICATION AND TROPICALIZATION OVER NON-ARCHIMEDEAN FIELDS 7

As an example, let us consider an elliptic curve E in Weierstrass form y2 = x3 + ax + b 2 over K. The aﬃne curve E has a natural embedding in AK . The intersection E0 of E 2 2 with Gm,K has a natural tropicalization in R , which is one of the trees depicted in the next ﬁgure.

Figure 1. Tropicalization of an elliptic curve in Weierstraß form. The left hand side shows the case 3v(a) ≥ 2v(b), the right hand side the case 3v(a) < 2v(b). The numbers on the edges indicate the tropical multiplicities. If no number is given, the multiplicity is one.

If however E is a Tate curve, then the analytic space Ean contains a circle, i.e. it has topological genus 1. Hence the Weierstrass tropicalization does not faithfully depict the topology of the analytic space. As stated before, if X is a smooth curve over K, the Berkovich space Xan is some kind of generalized graph, allowing infinite ramification along a dense set of points. In particular, an it makes sense to talk about its leaves. The space of non-leaves H0(X ) in the Berkovich analytification Xan admits a natural metric which is defined via semistable models, see [BPR13], section 5.3. Any tropicalization of X with respect to a closed embedding in a torus carries a natural metric which is locally given by lattice length on each edge (and globally by shortest paths). Another problem in the case of the Weierstraß tropicalizations from figure 1 is the fact that tropical multiplicites 6= 1 are present. This may indicate that the tropicalization map will not be isometric on a subgraph, see [BPR11], Corollary 5.9. It is explained in [BPR11], Theorem 6.2, how one can construct better tropicalizations of the Tate curve. Based on a detailed study of the structure of analytic curves there are the following two important comparison theorems in [BPR11]. Theorem 3.1 ([BPR11], Theorem 5.20). Let Γ be any finite subgraph of Xan for a smooth, complete curve X over K. Then there exists a closed immersion ϕ : X,→ Y into a toric −1 variety Y with dense torus T such that the restriction ϕ0 : X0 = X ∩ ϕ (T ) ,→ T an n induces a tropicalization map tropϕ0 : X0 → R which maps Γ homeomorphically and isometrically onto its image. 8 ANNETTE WERNER

Theorem 3.2 ([BPR11], Theorem 5.24). Consider a smooth curve X0 over K and a closed immersion ϕ : X0 ,→ T into a split torus with associated tropical variety 0 Tropϕ0 (X0). If Γ is a compact connected subset of Tropϕ0 (X0) which has tropical mul- an tiplicity one everywhere, then there exists a unique closed subset Γ in H0(X0 ) mapping homeomorphically onto Γ0, and this homeomorphism is in fact an isometry.

We will discuss higher-dimensional generalizations of these theorems in section 7.

4. Tropical Grassmannians

4.1. The setting. In view of the results [BPR11] for curves, it is a natural question whether they can be generalized to varieties of higher dimensions. As a ﬁrst example, the tropical Grassmannian of planes was studied in [CHW14]. In this section we discuss the main theorem of this paper.

For natural numbers d ≤ n we denote by Gr(d, n) the Grassmannian of d-dimensional subspaces of n-space. The tropical Grassmannian is defined as the tropicalization of (n)−1 Gr(d, n) with respect to the Plücker embedding ϕ : Gr(d, n) → P d . Recall that the Plücker embedding maps the point corresponding to the d-dimensional subspace W to n the line in d -space given by the d-th exterior power of W . During this section we will deviate from the exposition in section 2.3 and consider tropicalizations with respect to the negative valuation map. The choice between (min, +) and (max, +) tropical geometry is always a difficult one.

Hence our tropical projective space is (n)−1 (n) TP d = (R ∪ {−∞}) d r {(−∞,..., −∞)} /R(1,..., 1), and the tropical Grassmannian T Gr(d, n) = Tropϕ(Gr(d, n)) is deﬁned as the image of an an (n)−1 trop ◦ ϕ : Gr(d, n) → TP d , where trop is given by the map p 7→ (log |x0(p)|,..., log |x n (p)|) + R(1,..., 1) (d) on the analytic projective space, see (2.2).

Tropical Grassmannians were ﬁrst studied by Speyer and Sturmfels [SS04] who focused on n −1 n −1 −1 (d) (d) the toric part Gr0(d, n) = Gr(d, n) ∩ ϕ Gm,K which is embedded in the torus Gm,K 0 via ϕ. The tropicalization T Gr0(d, n) = Tropϕ(Gr0(d, n)) (which is called Gd,n in [SS04]) is a fan of dimension d(n − d) containing a linear space of dimension n − 1, see [SS04], section 3.

Moreover, if d = 2, Speyer and Sturmfels proved that T Gr0(2, n) can be identified with the space of phylogenetic trees, see [SS04], section 4. For our purposes, a phylogenetic tree on n leaves is a pair (T, ω), where T is a finite combinatorial tree with no degree-two ANALYTIFICATION AND TROPICALIZATION OVER NON-ARCHIMEDEAN FIELDS 9 vertices together with a labeling of its leaves in bijection with {1, . . . , n}, and ω is a real- valued function on the set of edges of T . The tree T is called the combinatorial type of the phylogenetic tree (T, ω). For every phylogenetic tree (T, ω) and all i 6= j in {1, . . . , n} we denote by xij the sum of the weights along the uniquely determined path from leaf i to leaf j. This tree-distance function satisfies the four-point-condition, which states that for all pairwise distinct indices i, j, k, l in {1, . . . , n} the maximum among

xij + xkl, xik + xjl, xil + xjk is attained at least twice.

4.2. A section of the tropicalization map. For the rest of this section we will always consider the case d = 2. In [CHW14], we investigate the full projective Grassmannian (n)−1 T Gr(2, n) = TropϕGr(2, n) in tropical projective space TP 2 . The main result of this paper is the following theorem: Theorem 4.1 ([CHW14], Theorem 1.1). There exists a continuous section σ : TGr(2, n) → Gr(2, n)an of the tropicalization map trop◦ϕan : Gr(2, n)an → TGr(2, n). Hence, the tropical Grassmannian TGr(2, n) is homeomorphic to a closed subset of the Berkovich analytic space Gr(2, n)an.

Note that we are not considering semistable models or their Berkovich skeleta in this approach. The first idea one might have is to look at the big open cells of the Grassmannians. Since d = 2, the coordinates of the ambient projective space are indexed by the two element subsets {i, j} of {1, . . . , n}. If i < j, we write pij for this coordinate, and we put pji = −pij. We intersect the Grassmannian with the standard open affine covering of projective space and get open affine subvarieties −1 Uij = Gr(2, n) ∩ ϕ {pij 6= 0}.

The affine Plücker coordinates are ukl = pkl/pij. Since the Plücker ideal is generated by the relations pijpkl − pikpjl + pilpjk = 0 for {i, j, k, l} running over the four-element n(n−2) subsets of {1, . . . , n}, the subvariety Uij can be identified with AK by means of the coordinates uik = pik/pij and ujk = pjk/pij for all k different from i and j.

n(n−2) an Recall that (AK ) is the set of all multiplicative seminorms on K[uik, ujk : k∈ / {i, j}] n(n−2) which extend the absolute value on K. As explained in section 2, the toric variety AK has a natural tropicalization. With the sign conventions in the present section, it is given by n(n−2) an 2(n−2) trop : (AK ) → (R ∪ {−∞}) p 7→ (log |uik(p)|, log |ujk(p)|)k∈{ / i,j} This induces the tropicalization map an 2(n−2) trop : Uij −→ (R ∪ {−∞}) . 10 ANNETTE WERNER

N an Now the tropicalization map on an affine space (AK ) has a natural section, which is given N N an N by the map δ :(R∪{−∞}) → (AK ) , mapping a point r = (r1, . . . , rN ) ∈ (R∪{−∞}) to the seminorm N X α Y (4.1) δ(r): K[x1, . . . , xN ] −→ ≥0 | cαx (δ(r))| = max{|cα| exp(riαi)}. R α N i=1 α∈N0 N Here we put exp((−∞)α) = 0 for all α ∈ N0. Note that the restriction of δ to R is a N N an N map from R to the torus (Gm) . We call the image of δ on (R ∪ {−∞}) the standard N an N an skeleton of (AK ) , and the image of δ|RN the standard skeleton of the torus (Gm) . However, the map δ for N = n(n−2) does not in general provide a section of the tropicalization map on the whole of Uij. Consider n ≥ 4, and let x = (xkl)kl be a point in the tropical Grassmannian TGr(2, n) which lies in the image of Uij under the tropicalization map. 2(n−2) Let ω be the projection to the affine coordinates ω = (xik −xij, xjk −xij)k∈{ / i,j} ∈ R . an Then δ(ω) is a point in the Berkovich space Uij with log |uik(δ(ω))| = xik − xij and log |ujk(δ(ω))| = xjk − xij. Since ukl := pkl/pij = uikujl − uilujk by the Plücker relations, the definition of δ gives us log |ukl(δ(ω))| = max{xik + xjl, xil + xjk} − 2xij. Hence δ can only provide a section of the tropicalization map if this maximum is equal to xkl + xij. This may fail if the labelled tree T has the wrong shape, e.g. if it looks like this:

The strategy for the proof of Theorem 4.1 is to compare the tropical Grassmannian with a standard skeleton of an affine space on a smaller piece of the tropicalization, namely on the part consisting of phylogenetic trees such that the underlying combinatorial tree has the right shape. If we want to make this precise, the definition of these smaller pieces is quite involved. A considerable part of the difficulties is due to the fact that we take into account the boundary strata of the Grassmannian in projective space. In order to explain the general strategy we will from now on restrict our attention to the torus part of the tropical Grassmannian. The general case can be found in [CHW14].

4.3. Sketch of proof in the dense torus orbit. In this section we explain the proof of Theorem 4.1 for the subset TGr0(2, n) of the tropical Grassmannian TGr(2, n). Recall that TGr0(2, n) is the tropicalization of the dense open subset Gr0(2, n) of the Grassmannian which is mapped to the torus via the Pl¨ucker map.

We fix a pair ij as above and work in the big open cell Uij. The coordinate ring Rij of Uij is a polynomial ring K[uik, ujk : k∈ / {i, j}] in 2(n − 2) variables. The other Plücker coordinates are expressed as ukl = uikujl − uilujk in Rij. Note that the affine variety Gr0(2, n) is contained in Uij. The coordinate ring of Gr0(2, n) is equal to the localization of Rij after the multiplicative subset generated by all ukl for {k, l}= 6 {i, j}. ANALYTIFICATION AND TROPICALIZATION OVER NON-ARCHIMEDEAN FIELDS 11

We also ﬁx a labelled tree T with n leaves 1, . . . , n, and arrange T as in the following ﬁgure

Figure 2. Arrangement of T with subtrees T1,...,Tr. Deﬁnition 4.2. Let be a partial order on the set {1, . . . , n} r {i, j}. We write k ≺ l if k l and k 6= l. Then has the cherry property on T with respect to i and j if the following conditions hold:

(i) Two leaves of diﬀerent subtrees Ta and Tb for a, b ∈ {1, . . . r} as in ﬁgure 2 cannot be compared by . (ii) The partial order restricts to a total order on the leaf set of each Ta, a = 1, . . . , r. (iii) If k ≺ l ≺ m, then either {k, l} or {l, m} is a cherry of the quartet {i, k, l, m}, i.e. the subtree given by this quartet of leaves contains a node which is adjacent to both elements of the cherry:

Figure 3. Cherry property

An induction argument shows the following lemma: Lemma 4.3 ([CHW14], Lemma 4.7). Fix a pair of indices i, j, and let T be a tree on n labelled leaves. Then, there exists a partial order on the set {i, . . . , n} r {i, j} that has the cherry property on T with respect to i and j.

For an example of the inductive construction of such a partial order see [CHW14], Figure 3.

The leaves in each subtree Ta are totally ordered, say as s1 ≺ s2 ≺ ... ≺ sp, if Ta contains p = p(a) leaves. We consider the variable set Ia = {uis1 , . . . , uisp }∪{ujs1 , us1s2 , . . . usp−1sp }. Then I = I1 ∪ ... ∪ Ir is a set of 2(n − 2) aﬃne Pl¨ucker coordinates of the form ukl ∈ Rij.

We can successively reconstruct the variables of the form ujl which are not contained in I as follows: In the tree Ta with leaves s1 ≺ s2 ≺ ... ≺ sp, we have us1s2 = uis1 ujs2 −uis2 ujs1 , hence u = u−1(u + u u ). js2 is1 s1s2 is2 js1 12 ANNETTE WERNER

The right hand side is an expression in the variables contained in the coordinate set

I with uis1 inverted. Now we use the relation us2s3 = uis2 ujs3 − uis3 ujs2 to express u = u−1(u + u u ). Plugging in the expression for u we can write u as a js3 is2 s2s3 is3 js2 js2 js3 −1 polynomial in the variables in I plus all uik . Proceeding by induction, we ﬁnd for all m 6= i, j that −1 ujm ∈ K[ukl : ukl ∈ I][uik : k 6= i, j] and hence −1 (4.2) K[ukl : ukl ∈ I] ⊂ Rij ⊂ K[ukl : ukl ∈ I][uik : k 6= i, j].

This shows that the variable set I generates the function ﬁeld Quot(Rij) of Uij. Recall −1 that the coordinate ring of Gr0(2, n) is K[Gr0(2, n)] = S Rij, where S is the multiplicative subset of Rij generated by all ukl for {kl}= 6 {ij}. By the previous result, K[Gr0(2, n)] ± is equal to the localization of the Laurent polynomial ring K[ukl : ukl ∈ I] after the multiplicative subset generated by all ukl expressed as Laurent polynomials in the coordinates contained in I.

Deﬁnition 4.4. Let CT be the cone in TGr0(2, n) whose interior corresponds to the phylogenetic trees with underlying tree T . Let x = (xkl)kl + R(1,..., 1) be a point in ij an CT ⊂ TGr0(2, n). We associate to it a point σT (x) in Gr0(2, n) , i.e. a multiplicative seminorm on the coordinate ring of Gr0(2, n), as follows. For every Laurent polynomial P α ± I f = α cαu ∈ K[ukl : ukl ∈ I] (where α runs over Z ) we put ij Y |f(σ (x))| = max |cα| exp αkl(xkl − xij) , where α = (αkl)u ∈I . T α kl ukl∈I ± This deﬁnes a multiplicative norm on K[ukl : {kl} ∈ I] which has a unique extension to a ij multiplicative norm on the localization K[Gr0(2, n)]. Let σT (x) be the resulting point in an the Berkovich space Gr0(2, n) .

Since for every f ∈ K[Gr0(2, n)] the evaluation map on CT given by ij x 7→ |f(σT (x))| is continuous, we have constructed a continuous map ij an σT : CT → Gr0(2, n) .

an an We want to show that it is a section of the tropicalization map trop ◦ ϕ : Gr0(2, n) → −1 TGr0(2, n) on trop (CT ), which amounts to checking that ij log |ukl(σT (x))| = xkl − xij for all {kl}= 6 {ij} and for all x ∈ CT . ij For variables ukl in I this is clear from the definition of σT . Hence it holds in particular for all indices of the form {ik} for k∈ / {i, j}. In order to check this fact for the other ANALYTIFICATION AND TROPICALIZATION OVER NON-ARCHIMEDEAN FIELDS 13 indices, recall the definition of I after Lemma 4.3. Since u = u−1(u + u u ), we js2 is1 s1s2 is2 js1 find ij log |ujs2 (σT (x))| = max{−xis1 + xs1s2 , −xis1 + xis2 + xjs1 − xij}.

Since s1 and s2 are in the same subtree (see figure 2), we find that xis1 +xjs2 = xis2 +xjs1 ≥ ij xij + xs1s2 , which implies log |ujs2 (σT (x))| = xjs2 − xij. Inductively, we can show in this way our claim for all indices of the form {jk}. If we consider an index of the form {kl} where k, l∈ / {i, j}, we have ukl = uikujl − uilujk. If k is a leaf in the subtree Ta and l is a leaf in the subtree Tb for a < b, we find xkl + xij = xil + xjk > xik + xjl. Hence the non- ij ij archimedean triangle inequality gives log |ukl(σT (x))| = log |(uilujk)(σT (x))| = xkl − xij. If k and l are leaves in the same subtree Ta, we may assume that k ≺ l. If k is the predecessor of l in the total ordering restricted to Ta, we are done, since then ukl is contained in I. If not, we let m be the predecessor of l, so that k ≺ m ≺ l. The Plücker relations give uimukl = uikuml + uilukm, hence −1 ukl = uim(uikuml + uilukm).

Note that all variables on the right hand side except possibly ukm are contained in I. ij Hence we can calculate log |ukl(σT (x))| by developing ukm in a Laurent series in the variables in I. The resulting Laurent series does not allow any cancellation between the variables. Now we can apply the cherry property for the ordering , which says that {k, m} or {m, l} is a cherry for the quartet {i, k, m, l}. Hence we have

xil + xkm ≤ xim + xkl = xik + xml or xik + xml ≤ xim + xkl = xil + xkm. In both cases, one can check directly that our claim holds.

ij ± In fact, the section σT has a more conceptual description. Let ψ : Gr0(2, n) → SpecK[ukl : 2(n−2) ukl ∈ I] = Gm be the open embedding induced by (4.2), and let Σ be the stan- 2(n−2) an dard skeleton of (Gm ) as in (4.1). It is contained in the open analytic subvariety an Gr0(2, n) , since it consists entirely of norms and does therefore not meet a closed sub- ij variety of strictly lower dimension. Then the map σT is the composition of the projection 2(n−2) from CT to the coordinates (xkl − xij)ukl∈I ∈ R = Σ followed by the inclusion of Σ an in Gr0(2, n) .

−1 Recall from section 2.3 that for every x ∈ TGr0(2, n) the preimage tropϕ (x) under the tropicalization map is the Berkovich spectrum M(Ax) of an aﬃnoid algebra Ax. It ij follows from our construction that σT (x) is the unique Shilov boundary point of Ax. We can formulate this fact explicitely as follows. ij Lemma 4.5. [[CHW14], lemma 4.17] For every x ∈ CT the seminorm σT (x) constructed above has the following maximality property: For every f in the coordinate ring −1 an K[Gr0(2, n)] and every seminorm p ∈ tropϕ (x) ⊂ Gr0(2, n) we have ij |f(p)| ≤ |f(σT (x))|. 14 ANNETTE WERNER

ij As an immediate consequence we see that σT (x) does not depend on the choice of the ij ij partial ordering or on the pair ij, and that σT (x) = σT 0 (x) if x is contained in the intersection CT ∩ CT 0 . Hence we can patch these maps together and get a well-deﬁned an section σ : TGr0(2, n) → Gr0(2, n) of the tropicalization map. It follows from the construction that it is continuous.

This proves theorem 4.1 on the dense torus orbit Gr0(2, n). In order to define a section of the tropicalization map also on the boundary strata of Gr(2, n), we follow the same strategy of constructing an index set I such that the associated Plücker variables generate the function field of the Grassmannian. This construction is more involved, see [CHW14], section 4 for details. The most important problem here is to find these local index sets in such a way that the section is also continuous when passing from one stratum of the Grassmannian for another. This continuity statement is shown in [CHW14], Theorem 4.19. Using the index set I, we may also calculate the initial degenerations of all points in the tropical Grassmannian and deduce that their tropical multiplicity is one everywhere. We will explain a general argument showing the existence of a section in this case in paragraph 7.2. In [DP14], Draisma and Postinghel give a different proof of Theorem 4.1. They work with the affine cone over the Grassmannian Gr(2, n), define the section on a suitable subset and use torus actions and tropicalized torus actions to move it around. Also in this approach a maximality statement such as Lemma 4.5 is used.

5. Skeleta of semistable pairs

In the next three sections, we give an overview of the results in [GRW14]. In order to compare polyhedral substructures of Berkovich spaces with tropicalizations, we start by generalizing Berkovich’s notion of skeleta. Such skeleta are induced by the incidence complexes of the special fibers of suitable models. We extend this notion by adding a horizontal divisor on the model. For the rest of this paper, we fix an algebraically closed ground field K which is complete with respect to a non-archimedean, non-trivial absolute value.

5.1. Integral affine structures. For curves, metrics play an important role in the comparison results between tropical and analytic varieties, as we have seen in section 3. The right way to generalize this to higher dimensions is to consider integral affine structures. Let M be a lattice in the finite-dimensional real vector space MR = M ⊗Z R, and let N = Hom(N, Z) its dual, which is a lattice in the dual space NR = N ⊗Z R of MR. We × denote the associated pairing by h , i : MR × NR → R. Recall that Γ = log |K |.

An integral Γ-affine polyhedron in NR is a subset of NR of the form ∆ = v ∈ NR | hui, vi + γi ≥ 0 for all i = 1, . . . , r ANALYTIFICATION AND TROPICALIZATION OVER NON-ARCHIMEDEAN FIELDS 15 for some u1, . . . , ur ∈ M and γ1, . . . , γr ∈ Γ. Any face of an integral Γ-affine polyhedron ∆ is again integral Γ-affine. An integral Γ-affine polyhedral complex in NR is a polyhedral complex whose faces are integral Γ-affine.

An integral Γ-affine function on NR is a function from NR to R which is of the form v 7→ hu, vi + γ for some u ∈ M and γ ∈ Γ. More generally, let M 0 be a second finitely generated free 0 0 0 abelian group and let N = Hom(M , Z). An integral Γ-affine map from NR to N is a ∗ 0 ∗ R 0 function of the form F = φ + v, where φ : M → M is a homomorphism, φ : NR → N 0 0 0 m R is the dual homomorphism extended to NR, and v ∈ N ⊗Z Γ. If N = M = Z and m F = (F1,...,Fm): NR → R is a function, then F is integral Γ-affine if and only if each coordinate Fi : NR → R is integral Γ-affine. An integral Γ-affine map from an integral Γ-affine polyhedron ∆ ⊂ N to N 0 is defined R R as the restriction to ∆ of an integral Γ-affine map N → N 0 . If ∆0 ⊂ N 0 is an integral R R R Γ-affine polyhedron then a function F : ∆ → ∆0 is integral Γ-affine if the composition ∆ → ∆0 ,→ N 0 is integral Γ-affine. We say that an integral Γ-affine map F : ∆ → N 0 is R R unimodular if F is injective and if the inverse map F (∆) → ∆ is integral Γ-affine. Note that F (∆) is an integral Γ-affine polyhedron in N 0 . R

5.2. Semistable pairs. Note that since our ground ﬁeld K is algebraically closed, the ring of integers K◦ is a valuation ring which is not discrete and not noetherian. We begin by decribing the building blocks of the polyhedral substructures of Berkovich spaces which lend themselves to comparison with tropicalizations.

Let 0 ≤ r ≤ d be natural numbers and consider the K◦-scheme ◦ S = Spec(K [x0, . . . , xd]/(x0 . . . xr − π)) for some π ∈ K◦ satisfying |π| < 1. Then S is a flat scheme over K◦ with smooth generic fiber. Its special fibre contains r + 1 irreducible components whose incidence complex is an r-dimensional simplex. Now fix a natural number s ≥ 0 such that r + s ≤ d, and consider the principal Cartier divisor H(s) = div(xr+1) + ... + div(xr+s) on S .

As explained in section 2.2, the K◦-scheme S gives rise to two analytic spaces in the following way. On the one hand, we have an associated admissible formal scheme ◦ (5.1) S = Spf K {x0, . . . , xd}/(x0 . . . xr − π) which we get by completing S with respect to a non-zero element in K of absolute value < 1. Its analytic generic ﬁber is M(AS ) for the aﬃnoid algebra

AS = K{x0, . . . , xd}/(x0 . . . xr − π).

It is a subset of the analytiﬁcation of the generic ﬁbre SK = Spec(K[x0, . . . , xd]/(x0 . . . xr− π)) of S . 16 ANNETTE WERNER

Now we look at the tropicalization map on M(AS ) which only takes into account the ﬁrst r + s + 1 coordinates, i.e. the map r+s+1 Val : M(AS )\H → R≥0 p 7→ (− log |x0(p)|,..., − log |xr+s(p)|), s where H is the support of the Cartier divisor induced by H(s). Its image is ∆(r, π)×R≥0, r+1 r+1 where ∆(r, π) = {(v0, . . . , vr) ∈ R≥0 : v0 + ... + vr = − log |π|} is a simplex in R .

s We deﬁne a continuous section ∆(r, π) × R≥0 → M(AS ) as follows. Note that the projection (x0, x1, . . . , xd) 7→ (x1, . . . , xd) induces an isomorphism from M(AS ) to the aﬃnoid subdomain B = {p : − log |x1(p)| − ... − log |xr(p)| ≤ − log |π|} of M(K{x1, . . . , xd}).

Similarly, the projection (v0, v1, . . . , vr+s) 7→ (v1, . . . , vr+s) induces a homeomorphism s r+s ∆(r, π) × R≥0 → Σ = {(v1, . . . , vr+s) ∈ R≥0 : v1 + ... + vr ≤ − log |π|}

For every v = (v1, . . . , vr+s) ∈ Σ there is a bounded multiplicative norm || ||v on K{x1, . . . , xd} which is deﬁned as follows: X I || aI x ||v = max{|aI | exp(−i1v1 − ... − ir+svr+s)}. I I=(i1,...,id)

It satisﬁes − log ||x1||v −...−log ||xr||v = v1 +...+vr ≤ − log |π|. Therefore the point || ||v is contained in the aﬃnoid domain B. Hence there is a uniquely determined continuous map s σ : ∆(r, π) × R≥0 → M(AS ) making the diagram s σ ∆(r, π) × R≥0 / M(AS )

v7→|| ||v Σ / B commutative. The map σ is by construction a section of the map Val. We deﬁne an S(S,H(s)) ⊂ M(AS ) ⊂ (SK ) as the image of σ and call it the skeleton of the pair (S,H(s)).

Now we consider schemes which étalelocally look like some S . Definition 5.1. A strictly semistable pair (X ,H) consists of an irreducible proper flat ◦ scheme X over the valuation ring K and a sum H = H1 + ··· + HS of effective Cartier divisors Hi on X such that X is covered by open subsets U which admit an étalemor- phism ◦ (5.2) ψ : U −→ S = Spec(K [x0, . . . , xd]/(x0 ··· xr − π)) × for some r ≤ d and π ∈ K with |π| < 1. We assume that each Hi has irreducible ∗ support and that the restriction of Hi to U is either trivial or defined by ψ (xj) for some j ∈ {r + 1, . . . , d}. ANALYTIFICATION AND TROPICALIZATION OVER NON-ARCHIMEDEAN FIELDS 17

This is a generalization of de Jong’s notion of a strictly semistable pair over a discrete valuation ring [dJ96], if we add the divisor of the special fiber to the horizontal divisor H. It is sometimes convenient to include only the horizontal divisor H as part of the data, since it is a Cartier divisor, whereas the special fiber of X may not be one. For a more detailed discussion of this issue see [GRW14], Proposition 4.17. If d = 1, i.e. in the case of curves, semistable models and therefore nice skeletons are always available after an extension of the ground field. In higher dimensions, we have to allow alterations in order to find semistable models, see [dJ96], Theorem 6.5.

5.3. Skeleta. Let (X ,H) be a semistable pair in the sense of definition 5.1. The generic fiber X of X is smooth of dimension d. Hence d is constant in every chart U , whereas the numbers r and s may vary with U . We denote the special fiber of X by Xs = X ⊗K◦ K˜ .

Denote by V1,...,VR the irreducible components of the special fiber Xs of X . The Cartier divisors Hi which are part of our data give rise to horizontal closed subschemes Hi of X , which are locally cut out by a defining equation of Hi. Putting Di = Vi for PR+S i = 1,...,R and Di+R = Hi, we get a Weil divisor D = i=1 Di on X . This gives rise to a stratification of X , where a stratum is defined as an irreducible component of T S a set of the form i∈I Di\ i/∈I Di for some I ⊂ {1,...,R + S}. We call any stratum contained in the special fiber Xs a vertical stratum and denote the set of all vertical strata by str(Xs,H).

Now we want to glue skeleta of local charts together in such a way that the faces of the resulting polyhedral complex are in bijective correspondence with the vertical strata in str(Xs,H). In order to achieve this, we may have to pass to a smaller covering in the category of admissible formal schemes.

Let (X ,H) be a strictly semistable pair with a covering as in Definition 5.1. We consider the induced formal open covering of the associated admissible formal scheme X which is defined by completion. Hence X is covered by formal open subsets U which admit an étalemorphism ψ : U → S to a formal scheme S as in (5.1).

It is shown in [GRW14], Proposition 4.1 that, after passing to a refinement, the formal étalecovering ψ : U → S −1 has the property that ψ {x0 = ... = xr+s = 0} is a vertical stratum S in the special fiber Xs such that for every vertical stratum T the following condition holds: The closure T of T in Xs meets Us if and only if S ⊂ T .

We deﬁne the skeleton of (U,H|U ) as the preimage of the skeleton of the standard pair: −1 S(U,H|U ) = ψ (S(S,H(s)). This is a subset of the analytic generic ﬁber Uη of U which does not meet the horizontal divisor H. 18 ANNETTE WERNER

It follows from results of Berkovich [Ber99] that the ´etalemap ψ actually induces a s homeomorphism between S(U,H|U ) and S(S,H(s)) ' ∆(π, r) × R≥0. Hence the map s Val ◦ ψ : S(U,H|U ) → ∆(π, r) × R≥0 is a homeomorphism.

It is shown in [GRW14], 4.5 that the skeleton S(U,H|U ) only depends on the minimal stratum S contained in the special ﬁber of U. Therefore we denote it by ∆S and call it the canonical polyhedron of S. The dimensions of S and of the canonical polyhedron ∆S are deﬁned in an obvious way and add up to d, see [GRW14], Proposition 4.10.

s As we have seen, ∆S is homeomorphic to ∆(π, r) × R≥0 for suitable data r, s and π as s above. We call ∆(π, r) the ﬁnite part and R≥0 the inﬁnite part of ∆S.

Figure 4. The canonical polyhedron ∆S in the case r = 0 and s = 2, in the case r = s = 1 and in the case r = 2 and s = 0 (from left to right).

As explained above, S 7→ ∆S is a bijective correspondence between faces ∆S of the skeleton S(X ,H) and vertical strata S induced by the divisor D in the special ﬁber Xs. Recall that D is given by the horizontal divisor H plus all irreducible components of X (which we also call vertical divisors). Note that a vertical stratum T satisﬁes S ⊂ T if and only if ∆T is a closed face of ∆S.

Now we can glue ∆S and ∆S0 along the union of all canonical polyhedra associated to vertical strata R such that R ⊃ S ∪ S0. In this way we define the skeleton of (X ,H) as the union of all ∆S for strata S in the special fiber Xs: [ S(X ,H) = ∆S. S We obtain a piecewise linear space S(X ,H) whose charts are integral Γ-affine polyhedra.

an The skeleton S(X ,H) is a closed subset of the analytic space X \HK , where we write HK for the generic ﬁber of the support of H on X . Using methods from [Ber99], one can an prove that it is in fact a strong deformation retract of X \HK . In [GRW14], Theorem 4.13 it is shown that the retraction map can be extended to a retraction map from Xan to a suitable compactiﬁcation of the skeleton.

If the horizontal divisor H = 0, then the skeleton S(X , 0) is equal to Berkovich’s skeleton of a semistable scheme, see [Ber99]. ANALYTIFICATION AND TROPICALIZATION OVER NON-ARCHIMEDEAN FIELDS 19

6. Functions on the skeleton

Throughout this section we ﬁx a strictly semistable pair (X ,H). We use the notation from the previous sections.

We want to show that for every non-zero rational function f on X such that the support of the divisor of f is contained in HK , the function − log |f| factors through a piecewise integral Γ-aﬃne map on the skeleton. Moreover, we will show a slope formula for this map. Theorem 6.1 ([GRW14], Proposition 5.2). Let f be a non-zero rational function on X such that the support of div(f) is containd in HK . We put U = X\HK and consider the function an F = − log |f| : U → R. Then F factors through the retraction map τ : U an → S(X ,H) to the skeleton:

F U an / : R

τ F | % S(X ,H) S(X ,H)

Moreover, the restriction of F to S(X ,H) is an integral Γ-aﬃne function on each canonical polyhedron.

The theorem is proved by considering the formal building blocks with distinguished strata and using a result of Gubler [Gu07], Proposition 2.11.

Recall that we denote the dimension of X by d. The slope formula for F is basically a balancing condition around each (d−1)-dimensional canonical polyhedron of the skeleton which involves slopes in the direction of all adjacent d-dimensional polyhedra.

Let ∆S be a d-dimensional canonical polyhedron of the skeleton S(X ,H) containing the (d − 1)-dimensional canonical polyhedron ∆T . Then the stratum S in the special ﬁber Xs is contained in the closure T (which is a curve), and it is obtained as a component of the intersection of T with one additional irreducible component of the divisor D.

This component is either vertical, i.e. a component of the special fiber, or horizontal, i.e. s given by some component of H. If it is vertical, then the finite part of ∆S ' ∆(r, π)×R≥0 (i.e. the simplex ∆(r, π)) is strictly larger than the finite part of ∆T . In this case we say that ∆S extends ∆T in a bounded direction. If the component is horizontal, then the s infinite part of ∆S (i.e. the product R≥0) is strictly larger than the infinite part of ∆T . In this case we say that ∆S extends ∆T in an unbounded direction. 20 ANNETTE WERNER

We deﬁne the bounded degree degb(∆T ) as the number of canonical polyhedral ∆S extending ∆T in a bounded direction. Similarly, we deﬁne the unbounded degree degu(∆T ) as the number of canonical polyhedra ∆S extending ∆T in an unbounded direction.

In the case d = 1, i.e. if X is a curve, all ∆T are vertices. A one-dimensional canonical polyhedron ∆S extending ∆T in a bounded direction is simply an edge of length v(π) > 0. In this case, the stratum S is a component of the intersection of two irreducible components in the special ﬁber Xs. A one-dimensional canonical polyhedron ∆S extending ∆T in an unbounded direction is a ray of the form R≥0. In this case, the stratum S is a component of the intersection of the irreducible component of the special ﬁber given by T with a horizontal component of the divisor. Hence the degree degb(∆T ) is the number of bounded edges in ∆T , and degu(∆T ) is the number of unbounded rays starting in the vertex ∆T .

Now we want to deﬁne multiplicities via intersection theory. Since X is not noetherian, the standard intersection theory tools from algebraic geometry are not available. However, on admissible formal schemes one can use analytic geometry to associate Weil divisors to Cartier divisors, and there is a reﬁned intersection product with Cartier divisors, see [Gu98], [Gu03] and Appendix A in [GRW14].

Definition 6.2. For every vertex u ∈ ∆T we denote by Vu the associated irreducible component of the special fiber Xs. We define an integer α(u, ∆T ) as follows:

If ∆T has zero-dimensional ﬁnite part {u}, we simply put α(u, ∆T ) = degb(∆T ).

If not, then T lies in at least two irreducible components of the special fiber Xs, which means that the finite part of T is a simplex ∆(r, π) of dimension at least one. If U is a suitable formal chart, it is shown in [GRW14], Proposition 4.17 that there exists a unique effective Cartier divisor Cu on U such that its Weil divisor is equal to v(π)(Vu ∩ Us). We define −α(u, ∆T ) as the intersection number (Cu.T ), which is equal to the degree of the pullback of the line bundle associated to Cu from U to T .

Note that Cartwright [Ca13] introduced the intersection numbers α(u, ∆T ) in the (noetherian) situation where K◦ is discrete valuation ring. He uses them to endow the compact skeleton S(X ) with the structure of a tropical complex, see [Ca13], Deﬁnition 1.1. This notion also involves a local Hodge condition which plays no role in our slope formula. The paper [Ca13] develops a theory of divisors on tropical complexes and investigates their relation to algebraic divisors.

Here we also need multiplicities for rays in ∆T , which are deﬁned as one-dimensional faces of the unbounded part of T .

Deﬁnition 6.3. Let Hr be the horizontal component corresponding to the ray r in ∆T . We put α(r, ∆T ) = −(Hr.T ), where we take the intersection product of T with the Cartier divisor Hr on X . ANALYTIFICATION AND TROPICALIZATION OVER NON-ARCHIMEDEAN FIELDS 21

Let F be a function on the skeleton S(X ,H), which is integral Γ-aﬃne on each canonical polyhedron. Now we are ready to deﬁne outgoing slopes of F on a (d − 1)-dimensional canonical polyhedron ∆T along a d-dimensional polyhedron ∆S.

Definition 6.4. Let ∆T be a (d − 1)-dimensional canonical polyhedron and let ∆S be a d-dimensional canonical polyhedron of S(X ,H) containing ∆T . We denote by ∆(r, π) the finite part of ∆S. Let F : ∆S → R be an integral Γ-affine function. i) If ∆S extends ∆T in a bounded direction, then there exists a unique vertex w of ∆S not contained in ∆T . We put   1 1 X slope(F ; ∆T , ∆S) = F (w) − α(u, ∆T )F (u) , v(π) degb(∆T ) u∈∆T where we sum over all vertices u in ∆T . ii) If ∆S extends ∆T in an unbounded direction, then there exists a unique ray s in ∆S not contained in ∆T , and we put 1 X slope(F ; ∆T , ∆S) = dsF − α(r, ∆T )drF, degu(∆T ) r∈∆T where we sum over all rays in ∆T . For any ray r we denote by drF the derivative of F along the primitive vector in the direction of r.

If X is a curve, ∆T = u is a vertex and ∆S is an edge with vertices u and w, then 1 slope(F ; ∆T , ∆S) = v(π) (F (w) − F (u)). If ∆S is a ray s starting in u, then we simply have slope(F ; ∆T , ∆S) = dsF . In higher dimensions, the definition is more involved, since 1 the naive slope v(π) (F (w) − F (u)) depends on the choice of a vertex u in ∆T . Therefore 1 P we define a replacement for a weighted midpoint in ∆T as α(u, ∆T )u. degb(∆T ) u∈∆T Note that this point does not necessarily lie in ∆T , since α(u, ∆T ) may be negative. We can now formulate the slope formula for skeleta. Theorem 6.5 ([GRW14], Theorem 6.9). Let f ∈ K(X)× be a non-zero rational function such that the support of div(f) is contained in HK . Let F : S(X ,H) → R be the restriction of the function − log |f| to the skeleton. Then F is continuous and integral Γ- affine on each canonical polyhedron of S(X ,H), and for all (d−1)-dimensional canonical polyhedra we have X slope(F ; ∆T , ∆S) = 0,

∆S ∆T where the sum runs over all d-dimensional canonical polyhedra ∆S containing ∆T .

If X is a curve, the slope formula basically says that the sum of all outgoing slopes along edges or rays in a ﬁxed vertex is zero. In this case, the slope formula is shown in [BPR13, 22 ANNETTE WERNER

Theorem 5.15]. It is a reformulation of the non-archimedean Poinca´re-Lelongformula proven in Thuillier’s thesis [Thu05], Proposition 3.3.15. The Poincar´e-Lelongformula is c c an equation of currents in the form dd log |f| = δdiv(f), where dd is a certain distribution- valued operator. This version of the slope formula was generalized to higher dimensions in the ground-breaking paper [Ch-Du12], where a theory of diﬀerential forms and currents on Berkovich spaces is developed. The approach of [Ch-Du12] uses tropical charts and does not rely on models or skeleta. In higher dimensions we see no direct relation to our slope formula.

7. Faithful tropicalizations

We will now investigate the relation between skeleta, which are polyhedral substructures of analytic varieties, and tropicalizations, which are polyhedral images of algebraic or analytic varieties.

7.1. Finding a faithful tropicalization for a skeleton. We start with a strictly semistable pair (X ,H) with skeleton S(X ,H) and generic fiber X. We consider ra- n tional maps f : X 99K Gm,K from X to a split torus. If U ⊂ X is a Zariski open n subvariety where f is defined, then f|U : U → Gm,K induces a tropicalization Tropf (U) an an n of U, which is defined as the image of the map trop ◦ f : U → R as in section 2.3. Note that the skeleton is contained in the analytification of every Zariski open subset of X, since it only contains norms on the function field of X. n Definition 7.1. A rational map f : X 99K Gm,K from X to a split torus is called a faithful tropicalization of the skeleton S(X ,H) if the following conditions hold: i) The map trop ◦ f an is injective on S(X ,H). ii) Each canonical polyhedron ∆S of S(X ,H) can be covered by finitely many integral Γ-affine polyhedra such that the restriction of trop ◦ f an to each of those polyhedra is a unimodular integral Γ-affine map.

For the deﬁnition of unimodular integral aﬃne maps see section 5.1. It is easy to see that a rational map which is unimodular on the skeleton stays unimodular if we enlarge it with more rational functions on X, see [GRW14], Lemma 9.3.

◦ If we look at a building block S = SpecK [x0, . . . , xd]/(x1 . . . xr − π) as in Deﬁnition s 5.1 with the local tropicalization ∆(r, π) × R≥0, we ﬁnd that the coordinate functions x0, . . . , xr+s induce a faithful tropicalization. Collecting the corresponding rational functions on X for all canonical polyhedra of the skeleton, we get a rational map on X which is locally unimodular on the skeleton. It is shown in the proof of Theorem 9.5 of [GRW14] how to enlarge this collection of rational function in order to ensure injectivity of the tropicalization map on the skeleton. In this way one can show the following result. ANALYTIFICATION AND TROPICALIZATION OVER NON-ARCHIMEDEAN FIELDS 23

Theorem 7.2 ([GRW14], Theorem 9.5). Let (X ,H) be a strictly semistable pair. Then there exists a collection of non-zero rational functions f1, . . . , fn on X such that the re- n sulting rational map f = (f1, . . . , fn): X 99K Gm,K is a faithful tropicalization of the skeleton S(X ,H).

7.2. Finding a copy of the tropicalization inside the analytic space. Let us now start with a given tropicalization of a very aﬃne K-variety U, i.e. with a closed immer- n sion ϕ : U,→ Gm,K of a variety U in a split torus. As in section 2.3 we consider the tropicalization map an an n tropϕ = trop ◦ ϕ : U → R .

−1 an Recall that for every point ω ∈ Tropϕ(U) the preimage tropϕ (ω) ⊂ U of the tropicalization map is the Berkovich spectrum of an aﬃnoid algebra Aω. Lemma 7.3 ([GRW14], Lemma 10.2). If the tropical multiplicity of ω is equal to one, then Aω contains a unique Shilov boundary point.

This lemma follows basically from [BPR11], Remark after Proposition 4.17, which shows how the preimage of tropicalization is related to initial degenerations.

Now we can show that on the locus of tropical multiplicity one there exists a natural section of the tropicalization map.

Theorem 7.4 ([GRW14], Theorem 10.7). Let Z ⊂ Tropϕ(U) be a subset such that the tropical multiplicity of every point in Z is equal to one. By Lemma 7.3 this implies that −1 for every ω ∈ Z the aﬃnoid space tropϕ (ω) has a unique Shilov boundary point which we denote by s(ω).

The map s : Z → U an, given by ω → s(ω), is continuous and a partial section of the tropicalization map, i.e. on Z we have tropϕ ◦ s = idZ . Hence the image s(Z) is a subset of U an which is homeomorphic to Z.

Moreover, if Z is contained in the closure of its interior in Tropϕ(U), then s is the unique continuous section of the tropicalization map on Z.

We give a sketch of the proof. It is enough to show that the section s is continuous and uniquely determined under our additional assumption. Since everything behaves nicely × under base change we may assume that the valuation map K → R>0 is surjective. Let n us first consider the case that U = Gm,K and ϕ the identity map. Then the section s is n the identification of the tropicalization R (which has multiplicity one everywhere) with n the skeleton of Gm,K as defined in (4.1). It is clear from the explicit description of s in (4.1) that it is continuous in this case.

0 n Moreover, if s is a different section of the tropicalization map in the case U = Gm,K which satisfies s(ω) 6= s0(ω), we find a Laurent polynomial f on which those two seminorms differ. 24 ANNETTE WERNER

Since s(ω) is the unique Shilov boundary point in the ﬁber of tropicalization, s0(ω) applied to f is strictly smaller than s(ω) applied to f. Since s(ω) is equal to s0(ω) on all monomials, it follows that the initial degeneration of f at ω cannot be a monomial. Therefore ω is contained in the tropical hypersurface Trop(f). A continuity argument shows that the same argument works in a small neighbourhood of ω. This is a contradiction since Trop(f) has codimension one. Therefore s is indeed uniquely determined if ϕ is the identity map.

n For general ϕ : U,→ Gm,K , where U has dimension d ≤ n, one shows that that there n d n d exists a linear map Z → Z with the following property: Let α : Gm,K → Gm,K be d the corresponding homomorphism of tori and consider ψ = α ◦ ϕ : U → Gm,K . Let d S(Gm,K ) be the skeleton of the torus as in (4.1). Then for all ω ∈ Z we have {s(ω)} = −1 −1 d −1 −1 d tropϕ (ω)∩ψ (S(Gm,K )). This implies that s(Z) = tropϕ (Z)∩ψ (S(Gm,K )) is closed, from which we can deduce continuity of s. Uniqueness follows from uniqueness in the torus case by composing the sections with ψ. Note that the preceeding theorem does not make any assumption on the existence of specific models. If we assume that (X ,H) is a semistable pair and U = X\HK , then the image of the section s defined in Theorem 7.4 is contained in the skeleton S(X ,H). This is shown in [GRW14], Proposition 10.9. The preceeding theorem treats the case of tropicalizations in tori. Let ϕ : X,→ Y be a closed embedding of X in a toric variety Y associated to the fan ∆. Then we may consider an an an the associated tropicalization Tropϕ(X) of X, i.e. the image of trop ◦ ϕ : X → Y → N ∆, where N ∆ is the associated partial compactification of N , see section 2.3. We can R R R apply Theorem 7.4 to all torus orbits. In this way, we get a section of the tropicalization map on the locus Z ⊂ Tropϕ(X) of tropical multiplicity one which is continuous on the intersection with each toric stratum. It is a natural question under which conditions this section is continuous on the whole of Z. This might shed new light on Theorem 4.1 for the tropical Grassmannian.

References

[BPR11] M. Baker, S. Payne, J. Rabinoff: Nonarchimedean geometry, tropicalization and metrics on curves. Preprint 2011. http://arxiv.org/abs/1104.0320v3. [BPR13] M. Baker, S. Payne, J. Rabinoff: On the structure of non-archimedean analytic curves. Con- temp. Math. 605. AMS 2013, 93-121. [BaRu10] M. Baker, R. Rumely: Potential theory and dynamics on the Berkovich projective line. Math- ematical Surveys and Monographs 159. AMS 2010. [Ber90] V. Berkovich: Spectral theory and analytic geometry over non-Archimedean fields. Mathemat- ical Surveys and Monographs 33. AMS 1990. [Ber99] V. Berkovich: Smooth p-adic spaces are locally contractible. Invent. Math. 137 (1999) 1-84. [Ber04] V. Berkovich: Smooth p-adic spaces are locally contractible. II. In: Geometric aspects of Dwork theory. De Gruyter 2004, 293-370. [Bo14] S. Bosch: Lectures on Formal and Rigid Geometry. Lecture Notes in Mathematics 2105. Springer 2014. [BGR84] S. Bosch, S. Güntzer, R. Remmert: Non-Archimedean Analysis. Springer 1984. ANALYTIFICATION AND TROPICALIZATION OVER NON-ARCHIMEDEAN FIELDS 25

[Ca13] D. Cartwright: Tropical Complexes. Preprint 2013. http://arxiv.org/abs/1308.3813. [CHW14] M. Cueto, M. Häbich, A. Werner: Faithful tropicalization of the Grassmannian of planes. Math. Ann. 360 (2014) 391-437. [Ch-Du12] A. Chambert-Loir, A. Ducros: Formes diffeérentielles réelles et courants sur les espaces de Berkovich. Preprint 2012. http://arxiv.org/abs/1204.6277. [Co08] B. Conrad: Several approaches to non-archimedean geometry. In: p-adic geometry. University Lecture Series 45. AMS 2008, 9-63. [Du07] A. Ducros: Espaces analytiques p-adiques au sens de Berkovich. SéminaireBourbaki 2005/06. Astérisque311, exposé958 (2007). [dJ96] J.A. de Jong: Smoothness, semi-stability and alterations. Publ. Math. IHES 83 (1996) 51-93. [DP14] J. Draisma, E. Postinghel: Faithful tropicalization and torus actions. Preprint 2014. http://arxiv.org/abs/1404.4715. [Du] A. Ducros: Espaces de Berkovich, squelettes et théoriedes modèles.Confluentes Math. 4 (2012) 1250007. [EKL06] M. Einsiedler, M. Kapranov, D. Lind: Non-Archimedean amoebas and tropical varieties. J. Reine Angew. Math. 601 (2006) 139-157. [FGP14] T. Foster, P. Gross, S. Payne: Limits of tropicalizations. Israel J. Math. 201 (2014) 835-846. [Gu98] W. Gubler: Local heights of subvarieties over non-Archimedean fields. J. Reine Angew. Math. 498 (1998) 61-113. [Gu03] W. Gubler: Local and canonical heights of subvarieties. Ann. Sc. Norm. Super. Pisa Cl. Sci. (5) 2 (2003) 711-760. [Gu07] W. Gubler: Tropical varieties for non-Archimedean analytic spaces. Invent Math. 169 (2007) 321-376. [Gu13] W. Gubler: A guide to tropicalizations. In: Algebraic and combinatorial aspects of tropical geometry. Contemp. Math. 589. AMS 2013, 125-189. [GRW14] W. Gubler, J. Rabinoff, A. Werner: Skeletons and Tropicalizations. Preprint 2014. http://arxiv.org/abs/1404.7044. [Ma-St] D. Maclagan, B. Sturmfels: Introduction to Tropical Geometry. AMS, to appear 2015. [Pay09] S. Payne: Analytification is the limit of tropicalization. Math. Res. Lett. 16 (2009) 543-556. [SS04] D. Speyer, B. Sturmfels: The tropical Grassmannian. Adv. Geom. 4 (2004) 389-411. [Tem] M. Temkin: Introduction to Berkovich analytic spaces. In: Berkovich Spaces and Applications. Lecture Notes in Mathematics 2119. Springer 2015, 3-67. [Thu05] A. Thuillier: Théorie du potentiel sur les courbes on géométrie analytique non-archimédienne. Applications à la théorie d’Arakelov. Thèse Rennes 2005. http://tel.ccsd.cnrs.fr/documents/archives0/00/01/09/90/index.html [Tyo12] I. Tyomkin: Tropical geometry and correspondence theorem via toric stacks. Math. Ann. 353 (2012) 945-995.

Institut f¨urMathematik Goethe-Universit¨atFrankfurt Robert-Mayer-Strasse 8 D- 60325 Frankfurt email: [email protected]